Original Article

Asian Cardiovascular & Thoracic Annals

2018, Vol. 26(1) 5–10
ß The Author(s) 2017
Pericardial effusion following cardiac Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
surgery. A single-center experience DOI: 10.1177/0218492317744902
journals.sagepub.com/home/aan

Hien Sinh Nguyen1, Hung Doan-Thai Nguyen2 and

Thang Duc Vu3

Abstract
Background: Pericardial effusion is still a common postoperative complication after open heart surgery with cardio-
pulmonary bypass. Pericardial effusion significantly prolongs the hospital stay and associated costs as well as affecting
overall outcomes after open heart surgery in Hanoi Heart Hospital, a tertiary hospital in Vietnam with an annual volume
of 1000 patients. This study aimed to investigate the clinical presentation, incidence, and risk factors of postoperative
pericardial effusion, which may ensure better prevention of pericardial effusion and improvement in surgical outcomes
after open heart surgery.
Methods: A cross-sectional study was performed on 1127 patients undergoing open heart surgery from January 2015 to
December 2015.
Results: Thirty-six (3.19%) patients developed pericardial effusion. Of these, 16 (44.4%) had cardiac tamponade.
Pericardial effusion occurred after valve procedures in 77.8% of cases. Pericardial effusion was detected after discharge
in 47.2% of cases at a mean time of 18.1  13.7 days. Univariate logistic regression analysis showed that age > 25 years,
body surface area 5 1.28 m2, preoperative liver dysfunction, New York Heart Association class III/IV, left ventricular end-
diastolic diameter z score 5 0.55, and postoperative anticoagulant use were associated with postoperative pericardial
effusion. Multivariate logistic regression analysis showed that left ventricular end-diastolic diameter z score 5 0.55 was
an independent risk factor for postoperative pericardial effusion.
Conclusions: Routine postoperative echocardiography is necessary to detect postoperative pericardial effusion.
Increased left ventricular end-diastolic dimension is an independent predictor of postoperative pericardial effusion.

Keywords
Cardiac surgical procedures, Cardiac tamponade, Echocardiography, Pericardial effusion, Postoperative complications

Introduction postoperative PE is unknown.6 PE significantly affects

Pericardial effusion (PE) is still a common postopera-
tive complication after open heart surgery with cardio-
pulmonary bypass. PE has been reported to be
associated with less-favorable or even critical clinical
outcomes.1–3 Postoperative surgical or coagulopathic
bleeding and exudative fluid due to heart failure are
common causes of postoperative PE. Fibrin deposition
after pericardiectomy is initially a good response to
secure hemostasis and promote healing, but it may
lead to subsequent PE due to fibrinolysis and secondary
bleeding.4,5 Several risk factors have been shown to be
related to postoperative PE, including the type of car-
diac procedure, anticoagulants, heart failure, and renal
failure. However, the cause of the majority of
hospital stay, associated costs, and overall outcomes
after open heart surgery in Hanoi Heart Hospital, a
tertiary hospital in Vietnam with an annual volume of
1000 patients. PE might be related to the standard of
care, surgical techniques, heart failure, or kidney fail-
ure. However, their roles in the development PE in
1
Hanoi Heart Hospital, Hanoi, Vietnam
2
Hospital 175, Ho Chi Minh City, Vietnam
3
Hospital 103, Vietnam Military Medical University, Hanoi, Vietnam
Corresponding author:
Nguyen Sinh Hien, Hanoi Heart Hospital, 92 Tran Hung Dao Street, Hoan
Kiem, Hanoi, Vietnam.
Email: nguyensinhhien@gmail.com
6 Asian Cardiovascular & Thoracic Annals 26(1)

Vietnamese patients have not been well studied for Table 1. Demographic characteristics and surgical
better management of this condition. This study interventions in 1127 patients.
aimed to investigate the clinical presentation, incidence,
Pericardial
and risk factors of postoperative PE, which may ensure effusion No effusion
better prevention of PE and improve surgical outcomes Variable (n ¼ 36) (n ¼ 1091)
after open heart surgery.
Mean age (years) 45.5  16.7 33.7  25.3
Male 20 (55.6%) 557 (51.1%)
Patients and methods Female 16 (44.4%) 534 (48.9%)
The study was approved by the institutional review Coronary bypass 0 76 (6.8%)
board of Hanoi Heart Hospital. We selected 1127 Valve surgery 28 (77.8%) 483 (44.3%)
patients aged from 5 days to 81 years from 1276 Valve surgery þ coronary 0 18 (1.6%)
patients undergoing open heart surgery from January bypass
2015 to December 2015. PE patients were excluded Congenital cardiac 6 (16.7%) 499 (45.7%)
from the study if they were not undergoing surgery surgery
under cardiopulmonary bypass, had PE due to other Aortic surgery 2 (5.6%) 15 (1.4%)
causes such as cancer, were diagnosed more than
90 days after surgery, or had cardiac tamponade requir-
ing reoperation within the first 3 postoperative days.
Patients who died after surgery for reasons unrelated
to PE were also excluded. Table 2. Characteristics of 36 patients with postoperative
Patients were diagnosed with PE based on the pres- pericardial effusion.
ence of symptomatic fluid accumulation in the pericar-
dial cavity that required hospitalization for observation Variable No. of patients
and treatment, either medical or surgical. Postoperative Time of diagnosis
echocardiography was performed after surgery, before
Before discharge 19 (52.85%)
discharge, at 2 weeks and every month after discharge.
After discharge 17 (47.2%)
This was a cross-sectional study. Data were analyzed
using SPSS version 22 software (SPSS, Inc., Chicago, Mean postoperative stay (days) 18.1  13.7
IL, USA), with the chi-square test for categorical data 1–7 4 (11.1%)
and the independent t test or Mann-Whitney test for 8–30 27 (75%)
numerical data. A p value < 0.05 was considered signifi- >30 5 (13.9%)
cant. Univariate and multivariate logistic regression
Cardiac tamponade 16 (44.4%)
analyses were used for assessment of risk factors.

Results
Thirty-six patients (3.19%) had postoperative PE.
Their demographic data and surgical details are listed Table 3. Clinical presentation of pericardial effusion in 36
in Table 1. The male:female ratio was 1.05:1. Most PE patients.
(75.0%) occurred within 3 weeks after surgery when the Variable No. of patients
patients had already been discharged, and 16 (44.4%)
presented with cardiac tamponade. The mean time Tiredness* 34 (97.1%)
from surgery to diagnosis of PE was 18.1  13.7 days Breathless* 27 (77.1%)
(Table 2). Clinical presentations of PE were nonspecific Chest discomfort* 26 (74.3%)
and mild (Table 3). Patients with cardiac tamponade Edema* 3 (8.6%)
presented with tachycardia only, and did not show Tachycardiay 21 (58.3%)
signs of hemodynamic instability. Diagnosis of Low systolic BP (<90 mm Hg) 7 (19.4%)
PE and cardiac tamponade relied on routine
Oliguria 2 (5.6%)
postoperative follow-up using echocardiography
Hepatomegaly 8 (22.2%)
(Table 4). Echocardiography showed a mean pericar-
Prominent jugular vein 1 (2.8%)
dial separation of 31.7  8.9 mm (range 10–62 mm),
with 91.6% of patients having a large PE (>20 mm), Dull heart sounds 1 (2.8%)
and 86.1% of PE were loculated behind the left ven- *n ¼ 35 (one patient was under mechanical ventilation with mild
tricle. The mean drainage volume was 426.4  189.9 mL sedation). yHeart rate > 100 beatsmin1. BP: blood pressure.
Nguyen et al. 7

Table 4. Electrocardiographic and echocardiographic findings in Table 5. Potential risk factors for pericardial effusion.
patients with pericardial effusion.
Pericardial
Variable No. of patients effusion No effusion
Risk factor (n ¼ 36) (n ¼ 1091)
Electrocardiography (n ¼ 31)
Low voltage* 23 (74.2%) Age 4 25 years 14 (40%) 91 (8.3%)
ST elevation 14 (45.2%) Age > 25 years 21 (60%) 1000 (91.7%)
Electrical alternans 8 (25.8%) BSA (m2) 1.42  0.28 1.15  0.51
Echocardiography BSA < 1.28 m2 4 (11.1%) 430 (39.4%)
Widest pericardial separation (mm) 31.7  8.9 BSA 5 1.28 m2 32 (88.9%) 661 (60.6%)
Location of pericardial effusion Elevated liver enzymes 5 (13.9%) 56 (5.1%)
Behind left ventricle 31 (86.1%) Renal failure 3 (9.1%) 75 (6.9%)
Generalized 5 (13.9%) (>18-years old)
Size of pericardial effusion Hypertension 5 (13.9%) 128 (11.7%)
Small (<10 mm) 1 (2.8%) Diabetes 1 (2.8%) 47 (4.3%)
Large (>20 mm) 35 (97.2%) NYHA class III/IV 6 (16.7%) 68 (6.2%)
Cardiac tamponade NYHA class I/II 30 (83.3%) 1023 (93.8%)
Right atrial compression 3 (6.7%) LVEDD z score < 0.55 5 (13.9%) 435 (39.9)
Right ventricular compression 15 (33.3%) LVEDD z score 5 0.55 31 (86.1%) 656 (60.1%)
Fibrin accumulation 12 (33.3%) Left ventricular ejection 62.6%  9.7% 63.9%  10.6%
fraction
*Sum of QRS amplitudes in limb leads < 0.5 mV and sum of QRS amp- Valve surgery 28 (77.8%) 483 (44.3%)
litudes in precordial leads < 1 mV.
Coronary artery 0 76 (7%)
bypass surgery
(range 50–1000 mL). PE occurred as a transudate in 9 Aortic surgery 2 (5.6%) 15 (1.4%)
(25%) patients and hemopericardium in 27 (75%). The Emergency surgery 1 (2.8%) 36 (3.3%)
presence of some potential risk factors of PE in the 2 Bypass time (min) 87.9  34.5 84.6  42.9
groups are listed in Table 5. Univariate logistic regres- Aortic crossclamp 66.5  30.8 60.9  33.9
sion analysis showed that postoperative PE was asso- time (min)
ciated with age > 25 years, body surface area Hypothermia 1 (2.8%) 21 (1.9)
(BSA) 5 1.28 m2, preoperative liver enzyme elevation, Preoperative anticoagulant 16 (44.4%) 409 (37.5%)
New York Heart Association (NYHA) class III/IV, left Postoperative anticoagulant 31 (86.1%) 612 (56.1%)
ventricular end-diastolic diameter (LVEDD) z
score 5 0.55, and the use of postoperative anticoagu- BSA: body surface area; LVEDD: left ventricular end-diastolic diameter;
lants (Table 6). Multivariate logistic regression analysis NYHA: New York Heart Association.
showed that only LVEDD z score 5 0.55 was asso-
ciated with postoperative PE (Table 7). Surgical inter- moderate and do not require any treatment.7–9 Only
vention was indicated for cardiac tamponade in 36 1%–2% of PE are severe, requiring close monitoring
(44.4%) patients, clot evacuation in 9 (25%), large PE and intervention.8,10,11 In the present study, PE in
in 6 (16.7%), or moderate PE unresponsive to conser- patients aged 5-days to 81-years undergoing various
vative treatment in 5 (13.9%). Colchicine 1 mg daily types of cardiac surgery were investigated. Our data
and ibuprofen 800 mg daily for 1–2 weeks were used show that PE occurred in 3.19% of patients, and
for prevention of recurrent PE. All patients were fol- 44.4% of these presented with cardiac tamponade on
lowed up after 2 weeks and then monthly for 3 months. echocardiography only. Similar to previous studies,1,9
Only 4 (11.1%) patients treated with fluid aspiration most (75%) PE in our study were diagnosed in the 3rd
had a recurrent PE at 23  7 days after treatment, week after surgery when a half of these patients had
and required a second intervention with either a peri- already been discharged. Only 11.1% of PE were diag-
cardial window (1 patient) or subxiphoid pericardiect- nosed during the 1st week after surgery when the
omy for drainage (3 patients). patients were still in hospital.
Clinical presentations of PE are nonspecific, includ-
ing lethargy, breathlessness, and chest discomfort
Discussion
(97.1%, 77.1%, and 74.3%, respectively).1,2,12 Less fre-
The reported incidence of PE after open heart surgery is quent symptoms are edema and hepatomegaly (8.6%
approximately 77%. The majority of cases are mild to and 22.2%, respectively). Classic signs of PE such as
8 Asian Cardiovascular & Thoracic Annals 26(1)

Table 6. Univariate analysis of predictive factors for (25.8%). Large PE (>20 mm) were common in our
postoperative pericardial effusion. study (91.6%), with a mean pericardial separation of
31.7  8.9 mm, and 86.1% loculated posterior to the left
Odds
Risk factor ratio 95%CI p value heart. Ashikhmina and colleagues1 reported a similar
pericardial separation of 30  12 mm (range 7–70 mm),
Male sex 1.2 0.6–2.3 0.595 but a lower rate of large PE (41%) and loculated PE
Age > 25 years 7.3 2.2–24.0 < 0.001 (64%), possibly due to differences between the 2 studies
BSA > 1.28 m2* 5.2 1.8–14.8 0.001 in terms of patient selection and the classification of
Elevated liver enzymes 2.97 1.11–7.94 0.041 PE. Pericardial blood clots were detected on echocardi-
Renal failure (>18 year-old) 1.36 0.4–4.6 0.408 ography in 33.3% of PE patients, which was greater
Hypertension 1.22 0.5–3.2 0.418 than the 10% reported by Ashikhmina and colleagues.1
Diabetes 0.6 0.1–4.9 0.545
Patients with large pericardial clots that caused cardiac
tamponade were treated with pericardiectomy for
NYHA class III/IV 3.02 1.2–7.5 0.026
decompression, clot evacuation, and hemostasis. The
LVEDD z score 4.1 1.6–10.7 0.002
largest clot observed in our study was 3  7 cm.
Valve surgery 4.41 1.99–9.75 < 0.001 The clinical presentation of PE did not correlate with
Coronary bypass 0.97 0.96–0.98 0.078 the amount of fluid accumulation, as shown by the
Aortic surgery 4.2 0.9–19.2 0.100 absence of signs of hemodynamic instability in patients
Emergency surgery 0.8 0.1–6.3 0.667 with large PE, including decreased systolic blood pres-
Hypothermia 1.5 0.2–11.5 0.505 sure and hemodynamic collapse. These patients only
Preoperative anticoagulant 1.33 0.7–2.6 0.397 showed tachycardia. The absence of critical symptoms
Postoperative anticoagulant 4.85 1.9–12.6 < 0.001 in PE patients with a large volume of fluid drained
(426.4  189.9 mL) might be explained by incomplete
*Cut-off point determined by receiver operator characteristic analysis.
closure of the pericardium, an enlarged pericardial
BSA: body surface area; CI: confidence interval; LVEDD: left ventricular
end-diastolic diameter; NYHA: New York Heart Association. cavity due to heart failure, and enhanced functional
compensation after surgical correction. It should be
noted that cardiac tamponade presented in 44.4% of
PE patients and was only detected by echocardiog-
Table 7. Multivariate analysis of predictive factors for raphy, highlighting the importance of routine post-
postoperative pericardial effusion. operative echocardiography for detection of PE.
Postoperative PE has been reported to be associated
Odds
Risk factor ratio 95%CI p value with several risk factors such as large BSA, hyperten-
sion, heart failure, kidney failure, type of surgery, use
Age > 25 years 2.8 0.4–18.1 0.283 of anticoagulants, and bypass time.1,9 Ashikhmina and
BSA 5 1.28 m2 1.7 0.4–6.9 0.482 colleagues1 found that BSA was an independent pre-
Liver failure 1.98 0.7–5.6 0.198 dictor of postoperative PE. Univariate analysis in our
NYHA class III/IV 2.5 0.9–6.5 0.065 study showed that BSA > 1.28 m2 was associated with a
LVEDD z score > 0.55 3.4 1.27–8.96 0.015 5.2-times higher risk of postoperative PE, whereas
Valve surgery 1.8 0.6–5.5 0.302 multivariate analysis showed no significant association,
Postoperative anticoagulant 0.9 0.2–3.9 0.853
possibly due to its dependence on the age variable.
Similar to that of Ashikhmina and colleagues,1 our
BSA: body surface area; CI: confidence interval; LVEDD: left ventricular data showed no association between sex and PE.
end-diastolic diameter; NYHA: New York Heart Association. However, Cheung and colleagues9 reported that
females had a higher risk of postoperative PE.9 Our
dull heart sounds and pericardial friction rub were not results showed an association between PE and pre-
present in our patients. Hemodynamic changes includ- operative heart failure, as represented by NYHA class
ing low systolic blood pressure, tachycardia, and oli- and LVEDD z score. Patients with significant cardiac
guria were seen in 63.9%, which was higher than the remodeling had a higher risk of PE. Preoperative
incidence of 42% reported by Ashikhmina and col- NYHA class III/IV was associated with a 3.02-times
leagues.1 However, these symptoms are not specific higher risk of PE. We also noted that patients with
for cardiac tamponade. Cardiac tamponade was LVEDD z scores > 0.55 had a 4.1-times higher risk of
detected by echocardiography in some patients, but it PE. To the best of our knowledge, this is the first time
was not severe enough to cause hemodynamic collapse. that LVEDD z score has been identified as a predictor
Other common signs of PE were low QRS voltage of postoperative PE. A mismatch between the heart and
(74.3%), ST elevation (45.2%), and electrical alternans the pericardial cavity after surgical repair may
Nguyen et al.
predispose the latter to fluid accumulation and PE.
Elevated liver enzymes may be seen in heart failure
patients due to decreased cardiac output, which may
result in impaired liver perfusion and ensuing hepato-
cellular necrosis. This association could explain the
2.97-times higher risk of PE in patients with elevated
liver enzymes on univariate analysis, but this associ-
ation was not significant on multivariate analysis.
Most (77.8%) PE patients were in the valve surgery
group. Univariate analysis showed that valve surgery
increased the risk of PE 4.41-times compared to other
cardiac surgery. This finding is similar to that of
Ashikhmina and colleagues,1 who noted that the
majority (83%) of their PE patients had undergone
valve surgery. Our patients having valve surgery
received an anti-vitamin K anticoagulant which
could be an independent risk factor for postoperative
PE. The greater proportion of valve surgery patient-
s > 25-years old might explain the 7.3-fold higher risk
of PE in this age group. None of our PE patients
underwent coronary artery bypass surgery; PE after
coronary artery bypass used to be a common compli-
cation, possibly due to oozing of blood from the inter-
nal mammary artery bed after harvesting.7 Recent
improvement in internal mammary artery harvesting
and hemostasis techniques, in addition to wide open-
ing of the left pleura during internal mammary artery
harvesting, may facilitate drainage of pericardial fluid
and prevent PE. Ascending aortic surgery has been
reported to be an independent risk factor for PE,13
possibly due to postoperative bleeding, an inflamma-
tory reaction, dissolution of blood clots, plasma weep-
ing of the vascular graft, or damage to the regional
lymphatic system.1,13 In our study, 2/15 ascending
aortic surgery patients had PE, but univariate analysis
showed no significant association, possibly due to the
small number of patients (15/1127) undergoing this
procedure.
The effect of cardiopulmonary bypass on inflamma-
tion, redistribution of fluid and electrolytes, hemolysis,
and blood dilution may increase the risk of PE. Bypass
and aortic crossclamp times have been shown to be
related to postoperative PE. It has been reported that
PE patients have longer crossclamp times, and every
10-min increment in bypass time increases the risk of
postoperative PE.1 However, data from our study
showed no significant association between bypass or
crossclamp times and PE, and there was no significant
difference in bypass time between the PE and non-PE
groups.
The use of anticoagulants may increase the risk of
postoperative PE. Malouf and colleagues8 reported that
incidences of large PE and cardiac tamponade were
significantly increased in patients receiving postopera-
tive anticoagulation. Our data indicate that
9
preoperative anticoagulants did not increase the risk
of postoperative PE, but patients receiving postopera-
tive anti-vitamin K preparations had a 4.84-times
higher risk of postoperative PE. However, multivariate
analysis showed that this effect is not substantial. There
was no significant difference between the PE and non-
PE groups with regard to left ventricular ejection frac-
tion, renal failure, hypertension, or diabetes, possibly
due to the young cohort of patients with congenital and
valve diseases in our study, in whom kidney failure,
hypertension and diabetes are less common.
We concluded that most cases of PE present at 2–3
weeks after surgery when patients have been dis-
charged. The symptoms of PE are mild and nonspecific,
and postoperative echocardiography is currently the
only effective means to diagnose and guide interven-
tion. Factors associated with increased risk of
postoperative PE are age > 25 years, large BSA, pre-
operative elevation of liver enzymes (>2 times the
baseline), NYHA class III/IV, LVEDD z score > 0.55,
valve surgery, and use of anticoagulants; LVEDD
z score > 0.55 is an independent predictor of postopera-
tive PE.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
Funding
The author(s) received no financial support for the research,
authorship, and/or publication of this article.
References
1. Ashikhmina EA, Schaff HV, Sinak LJ, et al. Pericardial
effusion after cardiac surgery: risk factors. patient profiles.
and contemporary management. Ann Thorac Surg 2010;
89: 112–118.
2. Ikaheimo MJ, Huikuri HV, Airaksinen KE, et al.
Pericardial effusion after cardiac surgery: incidence, rela-
tion to the type of surgery, antithrombotic therapy, and
early coronary bypass graft patency. Am Heart J 1988;
116: 97–102.
3. Stevenson LW, Child JS, Laks H and Kern L. Incidence
and significance of early pericardial effusions after cardiac
surgery. Am J Cardiol 1984; 54: 848–851.
4. Rother AL and Collard CD. Management of cardiothor-
acic surgical emergencies. In: Hensley FA Jr, Martin DE,
Gravlee GP (eds) A Practical Approach to Cardiac
Anesthesia, 3rd ed. Philadelphia: Lippincott Williams &
Wilkins, 2003.
5. Wernovsky G, Chung AC, Wessel MD and Ravishankar
C. Cardiac intensive care. In: Allen HD, Gutgesell HP,
Clark EB, Driscoll DJ (eds) Moss and Adams’ Heart
Disease in Infants, Children, and Adolescents, 7th ed.
10
Philadelphia: Wolters Kluwer Health/Lippincott Williams
& Wilkins, 2008, pp. 477–478.
6. Bryan AJ and Angelini GD. Pericardial effusion after open
heart surgery. Thorax 1990; 45: 655–656.
7. Pepi M, Muratori M, Barbier P, et al. Pericardial effu-
sion after cardiac surgery: incidence, site, size, and haemo-
dynamic consequences. Br Heart J 1994; 72: 327–331.
8. Malouf JF, Alam S, Gharzeddine W and Stefadouros MA.
The role of anticoagulation in the development of pericar-
dial effusion and late tamponade after cardiac surgery. Eur
Heart J 1993; 14: 1451–1457.
9. Cheung EW, Ho SA, Tang KK, Chau AK, Chiu CS and
Cheung YF. Pericardial effusion after open heart surgery
for congenital heart disease. Heart 2003; 89: 780–783.
Asian Cardiovascular & Thoracic Annals 26(1)
10. Kuvin JT, Harati NA, Pandian NG, Bojar RM and
Khabbaz KR. Postoperative cardiac tamponade in the
modern surgical era. Ann Thorac Surg 2002; 74: 1148–1153.
11. Kulik A, Rubens FD, Wells PS, et al. Early postoperative
anticoagulation after mechanical valve replacement: a
systematic review. Ann Thorac Surg 2006; 81: 770–781.
12. Price S, Prout J, Jaggar SI, Gibson DG and Pepper JR.
‘Tamponade’ following cardiac surgery: terminology
and echocardiography may both mislead. Eur J
Cardiothoracic Surg 2004; 26: 1156–1160.
13. Eryilmaz S, Emiroglu O, Eyileten Z, et al. Effect of pos-
terior pericardial drainage on the incidence of pericardial
effusion after ascending aortic surgery. J Thorac
Cardiovasc Surg 2006; 132: 27–31.